JP2000125206A - Fixed pattern noise correcting device and its correcting method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain an optimum fixed pattern noise without needing a special optical mechanism nor interrupting normal image pick up by making a correction data holding means update the output of an adder means as new correction data. SOLUTION: Output image data 6 from a subtracter 5 is inputted to an adder 8. The correction data of the current frame corresponding to each pixel in accordance with the timing of the data 6 is inputted to the adder 8. Correction data 11 to which the correction data is added by the adder 8 is inputted to a frame memory 10 and a multiplier 9. When the correction data 11 of the current frame is fetched, the correction data of the preceding frame is replaced with the correction data 11 in the frame 10. The correction data 11 inputted multiplier 9 is made fixed pattern noise(FPN) data 7 undergoing constant number multiple and is inputted to the substracter 5. Thus, an FPN correction circuit 4 integrates the error of output image data 6 to an input image data 3 and feedbacks it.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an infrared sensor that removes uneven brightness between pixels due to characteristic variations of each pixel of an infrared sensor and uneven brightness of an image due to unnecessary infrared light incident from outside the field of view. The present invention relates to a fixed pattern noise correction device and a correction method for improving image quality.

[0002]

2. Description of the Related Art Conventional FPN (Fixed Pattern)
Noise (fixed pattern noise) correction method is as follows. First, create a uniform infrared incident environment over the entire field of view. The image data obtained at this time is
The data is stored in the memory. By subtracting this FPN data from the input image data when actually taking an image, image data from which the FPN has been removed can be obtained. Here, in order to create a uniform infrared incident environment, there is a method of covering the field of view of the imaging device with a plate having a uniform temperature. Further, there is a method of providing an optical shutter inside the imaging device and closing the shutter when acquiring FPN data. Further, the lens of the optical system is moved so that infrared light emitted from the subject is not formed on the sensor surface. That is, there is a method of creating an extreme defocus state and acquiring FPN data at this time.

[0003] By the way, FPN in an infrared imaging device
Is changed depending on the temperature environment in the imaging apparatus, so that it is necessary to periodically update the FPN data. In particular, immediately after the power is turned on, the temperature inside the imaging apparatus changes greatly, and it is necessary to frequently update the FPN data in order to obtain a high-quality image.

[0004]

However, the conventional FP
In the N correction method, normal acquisition of imaging is forced during acquisition of FPN data. This is not a particular problem in an imaging device used indoors or an application that does not require real-time properties. However, in applications requiring a severe temperature environment, real-time performance, and instantaneous startup performance, it is necessary to perform normal imaging even during acquisition of FPN data. An example of such an application is an infrared imaging device mounted on a missile infrared image guidance device. In such an infrared imaging apparatus, in order to detect and track a target as far as possible, it is important to find a local brightest point of about several pixels to several tens of pixels from information of captured image data. . In this type of imaging apparatus, the relative position of the imaging space and the imaging apparatus constantly changes, and the obtained image data changes every moment. Therefore, it is necessary to frequently update the FPN data in order to obtain a high-quality image. is there.

Japanese Patent Application Laid-Open No. 6-181545 discloses a technique in which FPN data is sequentially updated to obtain optimal FPN data. However, a special optical mechanism for obtaining FPN data is disclosed. Is required.

The present invention has been made in view of such a situation, and does not require a special optical mechanism and can obtain optimum FPN data without interrupting normal imaging. An object of the present invention is to provide a noise correction device and a correction method thereof.

[0007]

According to a first aspect of the present invention, there is provided a fixed pattern noise correction apparatus for correcting a fixed pattern noise with respect to an analog input image signal, and holding the correction data. Correction data holding means, and addition means for adding correction data to output image data obtained by converting an input image signal into digital data.The correction data holding means updates the output of the addition means as new correction data. Features. Further, it is possible to include a multiplying unit for multiplying the output of the adding unit by a constant, and a subtracting unit for subtracting the output of the multiplying unit from the output image data. The multiplication unit multiplies the output of the D / A conversion unit by the D / A conversion unit. The subtraction unit outputs the output of the multiplication unit to the analog input image signal. It can be made to subtract. Further, the correction data holding means may be a frame memory for holding correction data for each pixel of the previous frame. A fixed pattern noise correction method according to claim 5, wherein the fixed pattern noise is corrected for an analog input image signal, and the input image signal is corrected to digital output image data. The first to add data
And a second step of updating the added output as new correction data. Also,
The first and second steps can include a step of multiplying the added output by a constant, and a step of subtracting the output multiplied by the output image data. The first and second steps include a step of D / A converting the added output, a step of multiplying the D / A converted output by a constant, and a step of multiplying the analog input image signal. Subtracting the output that has been obtained. In the fixed pattern noise correction device and the correction method according to the present invention, the correction data held in the correction data holding unit is updated as the output of the adding unit as new correction data, and the output image data or the input image signal is updated. Is subtracted from the output multiplied by a constant of the multiplication means.

[0008]

Embodiments of the present invention will be described below. (First Embodiment) FIG. 1 is a block diagram showing a first embodiment of a fixed pattern noise correction device of the present invention, and FIG. 2 is a diagram for explaining the operation of the fixed pattern noise correction device of FIG. It is a graph of.

In FIG. 1, the fixed pattern noise correction device includes an FPN correction circuit 4. FP
The N correction circuit 4 converts the output image data 6 obtained by performing FPN (Fixed Pattern Noise) correction on the output data from the A / D converter 2 into a signal processing circuit 14.
Output to the side. The A / D converter 2 converts the input image signal 1 into input image data 3. Input image signal 1
Is an analog signal from which a luminance value for each pixel captured by an infrared sensor (not shown) is sequentially sent out. The input image data 3 is a digital signal converted by the A / D converter 2.

The FPN correction circuit 4 includes a subtractor 5, an adder 8, a multiplier 9, and a frame memory 10. The subtracter 5 removes the FPN component of the input image data 3. The removed data becomes output image data 6. The adder 8 adds the correction data output from the frame memory 10 as the correction data holding means to the output image data 6. The frame memory 10 holds correction data for each frame of a specific pixel. This correction data is replaced with the correction data 11 of the current frame output from the adder 8 immediately after output to the adder 8. Thereby, the correction data for each frame of the specific pixel is updated.

The multiplier 9 multiplies the correction data 11 output from the adder 8 by a constant. FPN data 7 multiplied by a constant
Is output to the subtractor 5. The signal processing circuit 14 performs signal processing on the output image data 6 to which the offset data 13 has been added by the adder 12. The signal processing here is, for example, image processing such as spatial filter processing or signal conversion processing for output to a display device.

The fixed pattern noise correction device having such a configuration performs the following operation. First, an input image signal 1 from an infrared sensor (not shown) is converted into digital signal input image data 3 by an A / D converter 2. After the input image data 3 is subtracted by the FPN data 7 corresponding to each pixel in the subtracter 5 of the FPN correction circuit 4,
The output image data 6 is input to the adder 12. The output image data 6 input to the adder 12 is input to a signal processing circuit 14 after appropriate offset data 13 is added thereto, and is subjected to desired processing.

On the other hand, the output image data 6 from the subtractor 5
Is input to the adder 8. The correction data of the current frame corresponding to each pixel is input to the adder 8 in accordance with the timing of the output image data 6. The correction data 11 to which the correction data has been added in the adder 8 is input to the frame memory 10 and the multiplier 9. When the frame memory 10 receives the correction data 11 of the current frame, the correction data is replaced with the correction data of the previous frame. Multiplier 9
Are input to the subtractor 5 as the FPN data 7 multiplied by a constant.

Therefore, the FPN correction circuit 4 is configured to integrate an error of the output image data 6 with respect to the input image data 3 and feed it back. For this reason, the output image data 6 is controlled to always become zero with a certain response. The response at this time depends on the gain of the multiplier 9. When the gain is close to 0, only the low frequency components are attenuated because the response is low. Gain 1
When the value is close to, the responsiveness is high, so that high-frequency components are also attenuated.

This state will be described with reference to FIG. FIG.
Indicates the temporal change of the luminance value of each specific pixel for each frame. A symbol a indicates input pixel data of a specific pixel of the input image 3 for each frame. Symbol b indicates output pixel data of each specific frame of the output image data 6 for each frame. FIG. 2 shows a situation where the DC component or the low frequency component of the output pixel data b with respect to the input pixel data a is attenuated.

Further, the waveform of the output pixel data b shows the case where the gain of the multiplier 9 is set to 0.1. As shown in FIG. 2, the DC component or the low frequency component is attenuated with time, as can be seen by comparing the input pixel data a and the output pixel data b. As a result, only the high frequency component is output from the FPN correction circuit 4 as the output image data 6. In other words, the FPN correction circuit 4 operates like a kind of high-pass filter that attenuates only components that do not change with time in the luminance value of each pixel. Note that the output image data 6 output from the FPN correction circuit 4 has a negative value unlike normal image data because the DC component is cut off. Therefore, in the adder 12, the appropriate offset data 13
Are superimposed, output as the output image data 6 in a state where the negative value is eliminated, and input to the signal processing circuit 14.

As described above, in the first embodiment, it is possible to remove only a component that does not change with time in the luminance value of each pixel. Here, the components that do not change with time include those caused by variations in the characteristics of the infrared sensor and unnecessary infrared light. Therefore, without providing a special optical mechanism for acquiring FPN data, it is possible to remove the variation in the characteristics of the infrared sensor and the uneven brightness caused by unnecessary infrared light. In addition, since the FPN data 7 is data updated for each frame, even if the FPN changes due to a change in the surrounding temperature environment or the temperature environment in the apparatus, the FPN data 7 becomes optimal data corresponding to the change.
As a result, the optimal F
PN correction can be performed.

(Second Embodiment) FIG. 3 is a block diagram showing a second embodiment of the fixed pattern noise correction device of the present invention. In the drawings described below, the same reference numerals are given to portions common to FIG. In the second embodiment, the process of subtracting the FPN component is performed at the stage of an analog signal. The FPN correction circuit 4a in the second embodiment has a D / D for converting digital correction data output from the adder 8 into analog data.
An A converter 2a is provided. The input image signal 1 is an analog signal obtained by sequentially sending out a luminance signal output from an infrared sensor (not shown) for each pixel. This input image signal 1 is subtracted by an analog FPN signal 7a by a subtractor 5.

That is, the input image signal 1 is input to the adder 12 via the subtractor 5. In the adder 12, the offset signal 13a is added to the input image signal 1. The input image signal 1 that has passed through the adder 12 is input to the A / D converter 2. The output image data 6 converted into a digital signal by the A / D converter 2 is output to a signal processing circuit 14.
And input to the adder 8. The offset signal 13 a is added to the output image data 6 input to the signal processing circuit 14 by the adder 12.

In the adder 8, the correction data of the previous frame in the frame memory 10 is added to the output image data 6. The correction data 11 output from the adder 8 is input to the frame memory 10 and the D / A converter 2a. In the frame memory 10, as described in the first embodiment, the correction data of the previous frame is replaced with the correction data 11 output from the adder 8. The correction data 11 is D /
The signal is converted into an analog signal by the A converter 2a, multiplied by a constant by the multiplier 9, and then input to the subtractor 5 as an FPN signal 7a.

As described above, in the second embodiment, the FP
The subtraction of the FPN component by the N correction circuit 4a can be performed at the stage of an analog signal. In this case, similar to the first embodiment, no special optical mechanism is required, and normal imaging is interrupted. Optimum FPN data can be obtained without performing.

In the second embodiment, the APN component is subtracted at the stage of the analog signal, so that A
The dynamic range of the / D converter 2 can be used effectively, and the dynamic range of the output image data 6 can also be used effectively. That is, when the digital FPN data 7 is used as in the first embodiment, variations in the characteristics of the infrared sensor and uneven brightness due to unnecessary light are large, and the FPN data 7 itself may be large. In such a case, after the input image signal 1 is converted into a digital signal having a limited number of bits in the A / D converter 2 and the subtraction processing of the FPN data 7 is performed, the output image data after the FPN correction is obtained. 6 may be impaired. However, input image data 3
On the other hand, when the FPN data is sufficiently small, no problem occurs.

As described above, the fixed pattern noise correction apparatus of the present invention is suitable for applications requiring a severe temperature environment, real-time performance, and instantaneous startup performance.
In particular, a remarkable effect can be expected when applied to an imaging device mounted on a missile infrared image guidance device.

[0024]

As described above, according to the fixed pattern noise correction apparatus and the correction method according to the present invention, the correction data held in the correction data holding means is updated with the output of the adding means as new correction data. At the same time, the output obtained by multiplying the output image data or the input image signal by a constant of the multiplication means is subtracted, so that no special optical mechanism is required, and the optimum FPN data can be obtained without interrupting normal imaging. Can be obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of a fixed pattern noise correction device according to the present invention.

FIG. 2 is a graph for explaining the operation of the fixed pattern noise correction device of FIG.

FIG. 3 is a block diagram showing a first embodiment of a fixed pattern noise correction device of the present invention.

[Explanation of symbols]

 Reference Signs List 1 input image signal 2 A / D converter 2a D / A converter 3 input image data 4,4a FPN correction circuit 5 subtractor 6 output image data 7 FPN data 7a FPN signal 8,12 adder 9 multiplier 10 frame memory 11 correction data 14 signal processing circuit

Claims (7)

[Claims] 1. A fixed pattern noise correction device for correcting fixed pattern noise on an analog input image signal, comprising: correction data holding means for holding correction data; and an output obtained by converting the input image signal into a digital signal. A fixed pattern noise correction apparatus, comprising: an addition unit that adds the correction data to the image data, wherein the correction data holding unit updates an output of the addition unit as new correction data. 2. The fixed device according to claim 1, further comprising: a multiplying unit that multiplies an output of the adding unit by a constant, and a subtracting unit that subtracts an output of the multiplying unit from the output image data. Pattern noise correction device. 3. A D / A converter for performing D / A conversion on an output of said adding means.
/ A conversion means, wherein when the output of the D / A conversion means is multiplied by the multiplication means, the subtraction means subtracts the output of the multiplication means from the analog input image signal. Item 3. The fixed pattern noise correction device according to Item 2. 4. The fixed pattern noise correction device according to claim 1, wherein said correction data holding means is a frame memory and holds correction data for each pixel of a previous frame. 5. A fixed pattern noise correction method for correcting a fixed pattern noise on an analog input image signal, the method comprising adding correction data to output image data obtained by converting the input image signal into a digital signal. And a second step of updating the added output as new correction data. 6. The first and second steps include a step of multiplying the added output by a constant multiple, and a step of subtracting the output multiplied from the output image data. 6. The method according to claim 5, wherein
3. The fixed pattern noise correction method described in 1. 7. The first and second steps include a step of D / A converting the added output; a step of multiplying the D / A converted output by a constant multiple; The method according to claim 5, further comprising the step of: subtracting the multiplied output from the input image signal.